Databases Reference
In-Depth Information
19
Video Compression
19.1 Overview
Video compression can be viewed as image compression with a temporal com-
ponent since video consists of a time sequence of images. From this point of
view, the only “new” technique introduced in this chapter is a strategy to take
advantage of this temporal correlation. However, there are different situations
in which video compression becomes necessary, each requiring a solution spe-
cific to its peculiar conditions. In this chapter we briefly look at video compression algorithms
and standards developed for different video communications applications.
19.2 Introduction
Of all the different sources of data, perhaps the one that produces the largest amount of data is
video. Consider a video sequence generated using the CCIR 601 format (Section
19.4
). Each
image frame is made up of more than a quarter million pixels. At the rate of 30 frames per
second and 16 bits per pixel, this corresponds to a data rate of about 21 Mbytes or 168 Mbits
per second. This is certainly a change from the data rates of 2.4, 4.8, and 16 kbits per second
that are the targets for speech coding systems discussed in Chapter 18.
Video compression can be viewed as the compression of a sequence of images; in other
words, image compressionwith a temporal component. This is essentially the approachwe will
take in this chapter. However, there are limitations to this approach. We do not perceive motion
video in the same manner as we perceive still images. Motion video may mask coding artifacts
